Performance Improvement of Genetic Algorithm used for Multiobjective Optimization


The research work is aimed at the performance enhancement of Real-Coded Genetic Algorithms used for multi-objective optimization. Study started with identifying the potential areas in Multi-objective Genetic Algorithms (MOGAs) where modification might enhance their performance. Operators play very important role in the functioning of GAs. In GAs the roles of selection and recombination (or crossover) operators are very well defined. Selection operator controls the direction of a search and the recombination operator generates new vistas for the search. The efficacy of GAs on a particular problem hinges quite strongly on the degree of exploration and exploitation of search space by the recombination operator and the direction of search in the search space set by the selection operator. GA uses simple selection operators for solving single objective optimization problems (SOPs) but special selection operators are needed to solve MOPs. The study investigates the behavior of several selection schemes and proposes special scheme for performance enhancement of MOGA and used them in new MOGA framework.